Yong,

Thanks for the information. The HA aspect of StarlingX looks solid.

My question is w.r.t. an entire cluster/site failure... is there a way to recover data in such scenarios either by means of a snapshot or bringing up a backup copy as the live one in case of a site disaster?

Regards
Abhijit

On Fri, Mar 27, 2020 at 6:20 AM Hu, Yong <yong.hu@intel.com> wrote:

Hi Abhijit,

Could you describe a bit more about the disaster recovery? What are your use cases?

We have some designs about the high availability, under several circumstances:

  1. If some “worker” (compute) node goes down, the workloads (VMs) or K8S pods could be migrated to other available worker nodes.
  2. If one of storage node goes down, the data in Ceph cluster will keep uncorrupted, thanks to the replications in other storage nodes.
  3. If one controller goes down, the other one will take over the active responsibility, to sustain the system working, though with fault warnings. However, if both controllers are shutdown, the StarlingX cluster will not stand it anymore.

 

Mostly these are “software” features. With regard to the HW, it relies on your servers in my opinion.

 

Regards,

Yong

From: Abhijit Deodhar <abhideodhar@gmail.com>
Date: Thursday, March 26, 2020 at 11:26 PM
To: "starlingx-discuss@lists.starlingx.io" <starlingx-discuss@lists.starlingx.io>
Subject: [Starlingx-discuss] Disaster recovery in StarlingX

 

Does StarlingX have a disaster recovery feature between two StarlingX clusters? I couldn't find info on it in the documentation...any plans of adding it in the future?  

 

Regards

Abhijit